The major goal of this project is to develop and exploit tractable, preclinical mouse models of malignant melanoma that recapitulate the clinical progression of surgical removal of primary tumors, treatment with adjuvant/neoadjuvant therapy and clinical recurrence at distant sites. We see a translational opportunity to devise a novel strategy to prevent the clinical recurrence of metastatic disease. We remain convinced that properly employed, relevant mouse models can improve current ICB approaches by 1) discovering reliable predictive biomarkers for patient stratification; 2) determining combination treatments that enhance ICB efficacy; and 3) identifying mechanisms of resistance and new targets to overcome them. The models are being designed to incorporate normal immune system function to maintain an appropriate tumor microenvironment and provide superior tumor-host interactions, an approach that is significantly more likely to yield biologically and clinically relevant data. Our focus will be on immunotherapy, which cannot be studied in human PDXs at this point as mice bearing a fully functional humanized immune system remain a future goal. Since metastasis often targets internal organs and the timing of its recurrence may vary greatly, the preclinical model will allow for the non-invasive, long-term monitoring of disease progression within the immunocompetent mouse. To meet these criteria, we have optically labeled melanoma tissue that was never adapted to cell culture and labeled using a high-titer lentivirus encoding a luciferase/GFP fusion reporter protein, developed by this laboratory in collaboration with Dom Esposito at the Protein Expression Laboratory, Advanced Technology Program (Day et al., Pig Cell Melanoma Res 22:283-295, 2009). However, we have noted that even low expression of the xenogeneic reporters GFP and/or luciferase can render growth of labeled metastatic tumors unpredictable in syngeneic mice and severely limit the utility of any immunocompetent preclinical model. We have therefore developed transgenic mice in both FVB/N and C57BL/6 backgrounds that express the Luc/GFP fusion gene away from the trunk where most metastasis is observed (in this case the anterior pituitary through use of a rat growth hormone promoter) (Day et al., PLoS One. 2014 Nov 4;9(11):e109956). These glowing head mice are thus pre-tolerized to both foreign markers, and we note that clearly labeled metastatic lesions arise much more regularly, consistently, and with a stronger signal in the lungs of inoculated, resected host mice. These mice are currently being employed in our ever-improving preclinical models (Day et al., PLoS One, 2014), and are available for purchase from Jackson Labs. The primary goal now is to combine all the tools and reagents we have developed over the last few years to study the mechanisms by which various melanoma subtypes recur at metastatic sites when patients fail targeted and immunotherapeutic drug treatment. This extensive preclinical effort is being performed in collaboration with CAPR, and in collaboration with Drs. Nick Restifo (CCR), Jedd Wolchok (MSKCC) and Jennifer Wargo (MD Anderson). We have developed and characterized 4 different melanoma models (M1 through M4) that vary in terms of oncogenic drivers (e.g., mutant BRAF and CDK4) and initiating carcinogens (e.g., UV, DMBA). A wildtype BRAF/NRAS/NF1 metastatic melanoma model (so-called triple negative) has been developed using our UV-initiated HGF transgenic mouse. Metastatic BRAF mutant melanomas have also been generated and are being fully characterized and incorporated into our studies. Preclinical trials with immune checkpoint inhibitors will be the focus of our work, and studies using both anti-CTLA-4 and, in collaboration with MedImmune, anti-PD-L1) are already underway and providing novel insights into why some melanomas but not others respond to immune checkpoint inhibitors. To evaluate the response of our models to ICB, we implanted GDAs or their derived cell lines from each model into syngeneic C57BL/6 mice and treated them with either anti-CTLA-4 or isotype control antibodies. Neither tumor growth nor survival were affected by anti-CTLA-4 in M1 and M2. By contrast, 20-30% of M3 and 40% of M4 melanomas treated with anti-CTLA-4 showed impaired tumor growth or shrinkage, resulting in significantly improved survival for M4-bearing mice. Overall, our genetically and phenotypically distinct 4-model panel recapitulates the diversity and broad spectrum of human responses to anti-CTLA-4, from resistant M1 and M2 to partially and highly sensitive M3 and M4, respectively, and offer a valuable new platform for preclinical and mechanistic studies Thus, the responses in our 4 melanoma models to anti-CTLA-4 and anti-PD-L1 seem appropriate to their clinical counterparts. Computational T cell dysfunction and exclusion signatures (Tumor Immune Dysfunction and Exclusion, TIDE) were recently reported to predict anti-CTLA-4 and anti-PD-1 responses in melanoma patients. By applying this tool to the gene expression data from our models, 100% of the untreated M1 and M2 melanomas were predicted as non-responders and 75% of M3 and M4 as responders. Moreover, M1 tumors exhibited high dysfunction scores, while M2 had increased exclusion scores. Our data also suggest that neoantigens generated by UV exposure and other sources are, at least in part, responsible for responses of melanoma tumors to immune checkpoint inhibitors. A number of approaches, including in vivo screens, vaccines, and computational assessment, are underway to identify the neoantigens that are responsible. We are also studying how the immune/inflammatory microenvironment influences the response to immune checkpoint inhibitors, as well as the role of melanoma clonality and heterogeneity. Recently, we have developed several highly metastatic melanomas and their derived cell lines. These will be used to study the efficacy of neoadjuvant therapy in immune checkpoint response, which is being introduced in the clinic now. Moreover, we have developed new metastatic melanoma models that home to the brain or bone, which is a major unmet need. Studies are underway to determine why metastases successfully and specifically colonize at these two sites.